Abstract: Aiming at the problems of low enthusiasm among voting nodes, bribery from malicious nodes, and vulnerability of sequential block generation by agent nodes in the Delegated Proof of Stake (DPoS) consensus process, an optimized DPoS consensus algorithm based on evolutionary game theory (EG-DPoS) is proposed. Firstly, the credit mechanism is introduced to construct the node voting incentive model, and the credit value reward is given according to the voting situation of nodes, which effectively improves the voting enthusiasm of nodes. Secondly, based on the strategy of evolutionary game, a behavior reward and punishment mechanism is formulated, which presets the corresponding revenue functions for different behavior strategies of nodes in the voting and election stage and implements rewards and punishments, so as to curb the bribery and collusion behavior of malicious nodes and ensure the stability and fairness of the system. Finally, the proportional coefficient of credit value and voting weight in the process of agent node election is balanced to reduce the oligopoly phenomenon caused by nodes with high credit value, while the roulette selection algorithm is used to randomize the block generation order of agent nodes, to avoid nodes being attacked during block generation and improves the security of the system. Simulated results show that compared with DPoS algorithm, EG-DPoS algorithm reduces the average delay by 36.83%, increases the average throughput by 19.44%, and improves the ratio of voting nodes to total nodes by approximately 42%. Due to the voting incentive mechanism within EG-DPoS and the fixed voting time, along with the influence of evolutionary game strategies, nodes will exhibit more secure and efficient behavior as the system operates. This leads to higher block generation and consensus efficiency for agent nodes, thereby reducing delay while improving throughput and enthusiam of voting nodes. Additionally, Compared compared with other typical DPoS improved algorithms, EG-DPoS also has obvious performance advantages.

Key words: Evolutionary game, DPoS, Consensus algorithm, Blockchain, Credit incentive

摘要： 针对委托权益证明算法(DPoS)共识过程中，投票节点积极性不高、恶意节点贿赂拉票和代理节点按序出块易被攻击的问题，提出了一种基于演化博弈的DPoS优化共识算法(EG-DPoS)。首先，引入信用机制构建节点投票激励模型，根据节点的投票情况来给予信用值奖励，有效提高了节点的投票积极性；其次，基于演化博弈的策略制定了一种行为奖惩机制，对投票选举阶段中各节点的不同行为策略预设对应的收益函数并实施奖惩，以此来遏制恶意节点的贿赂合谋行为，保证了系统的稳定性和公平性；最后，平衡了代理节点选举过程中信用值和投票权重的比例系数，以减少高信用值节点造成的寡头现象，同时利用轮盘选择算法打乱代理节点的出块顺序，避免节点在出块过程中被攻击，提高了系统的安全性。仿真实验结果表明，与DPoS算法相比，EG-DPoS的平均时延降低了36.83%，平均吞吐量提高了19.44%，且参与投票的节点数与总节点数的比值提升约42%，。这是由于EG-DPoS内存在投票激励机制和设定了固定的投票时间，以及在演化博弈策略的作用下，节点会随系统的运行表现得更加安全高效，使得代理节点的出块效率和共识效率更高，因此能在降低时延的同时提升吞吐量和投票节点的积极性，并且与其他典型DPoS改进算法相比，EG-DPoS也具有明显的性能优势。

关键词: 演化博弈, 委托权益证明, 共识算法, 区块链, 信用激励